4.1.1 Novel candidate proteins as substrates of Nedd4-1 or Nedd4-2 are identified in a physiological substrate screening in mouse brain
There have been a number of biochemical and proteomics studies to identify putative substrates of Nedd4-1 and Nedd4-2 in vitro. For instance, Nedd4-1 was purified and identified as an E3 ligase for phosphatase and tensin homolog (PTEN) in HeLa cell lysates (Wang et al., 2007). Even though Nedd4-1 has been reported to regulate the expression level, phosphatase activity, and localization of PTEN (Christie et al., 2012; Trotman et al., 2007), subsequent studies have produced different results, indicating that the protein level and localization of PTEN is not changed in Nedd4-1 knockout and knockdown cells (Cao et al., 2008; Fouladkou et al., 2008). In addition, several studies have proposed different E3 ligases, including WWP2, XIAP, CHIP to regulate PTEN (Ahmed et al., 2012; Maddika et al., 2011; Van Themsche et al., 2009). In a recent study, it has been reported that PTEN is not a downstream target of Nedd4-1 but an upstream regulator of Nedd4-1, acting by suppressing its translation through the MTORC1-PI3K pathway (Hsia et al., 2014). Therefore, in vitro biochemical assays and proteomics approaches are not always sufficient to reliable identify substrate proteins for E3 ligases.
In the present study, we tackled this issue by using the brain specific Nedd4-1; Nedd4-2 double conditional knockout mice line, which is probably one of the most stringent tools available to identify substrates of these proteins. Given the fact that Nedd4-1 and Nedd4-2 localize to synapses and conjugate K63-linked polyubiquitin chains, which have been shown to regulate the endocytosis of substrate proteins (Kawabe et al., 2010; Maspero et al., 2013;
Scudder et al., 2014), we analyzed synaptic membrane fractions from brain tissues, which enabled us to avoid cell type specific and context dependent artifacts in substrate identification.
Our data show that with this method we can also detect proteins on the perisynaptic end-feet of astrocytes at tripartite synapses. Out of five putative substrate proteins identified to be upregulated by iTRAQ mass spectrometry, we focused on two astrocytic proteins, Kir4.1 and Connexin-43, and one postsynaptic neuronal protein, Prr7. The other two candidate proteins identified were mitochondrial proteins, NADH dehydrogenase iron-sulfur protein 6 (Ndufs6) and CDGSH iron-sulfur domain containing protein 1 (Cisd1), and not followed-up.
74 Identification of mitochondrial proteins in SM3 fractions by proteomic screening might be due to inevitable slight contaminations of sample by synaptic mitochondria. However, purified SM3 fractions were highly depleted of synaptic cytoplasmic content (Figure 3-2), and most of the peptides identified by mass spectrometry were transmembrane or membrane anchored proteins.
Although we did not test Ndufs6 and Cisd1 in our study, Nedd4-1 or Nedd4-2 might specifically regulate these mitochondrial proteins since other mitochondrial proteins were not upregulated in Nedd4-1/2 bDKO SM3 fractions. The Western blotting-based validation of the upregulated levels of Kir4.1, Connexin-43 and Prr7, along with the finding that all these proteins are ubiquitinated by Nedd4-1 and Nedd4-2 (Figure 3-3, 3-5, 3-11), show that all three proteins we tested are prominent substrates of Nedd4-1 or Nedd4-2 E3 ligases. This shows that our approach of combining iTRAQ quantitative mass spectrometry with synaptic membrane purification in order to identify putative substrate proteins of E3 ligases is highly reliable. The high accuracy in the identification of substrate proteins in our study is due to several critical steps during the analysis. Firstly, the purification of SM3 fractions provides a rather low complexity sample for mass spectrometry, which enabled us to obtain more reliable and accurate data in the quantitative analysis. Secondly, the automated in-gel digestion and iTRAQ labelling methods minimize the variability in protein digestion and labelling efficiency. Gel-based labelling methods also had less complexity of samples for mass spectrometry analyses since protein bands are separated before trypsin digestion and iTRAQ labelling (Schmidt et al., 2013). Thirdly, we limited the identification to substrates that were identified by at least five peptides in order to dilute signal artifacts due to the biased labelling efficiencies of different iTRAQ labels. Lastly, we performed complementary experiments in mass spectrometric analysis in term of labels used for control and knockout SM3 fractions.
4.1.2 Astrocytic Nedd4-1 and Nedd4-2 are important for the regulation of Kir4.1 and Connexin-43
Nedd4-1 and Nedd4-2 were first identified in a screen to identify developmentally downregulated genes in mouse brain (Kumar et al., 1992). Expression of 1 and Nedd4-2 is not only regulated in a development-dependent manner but also spatially, so that Nedd4-1 is expressed ubiquitously in several tissues such as brain, muscle, liver and kidney while Nedd4-2 expression seems to be more restricted in brain, liver and kidney (Anan et al., 1998; Donovan
75 and Poronnik, 2013; Kumar et al., 1997). In a number of studies, Nedd4-1 and Nedd4-2 were proposed to be involved in neurite outgrowth, neuronal cell fate determination, and neuronal cell survival. In the context of neurite outgrowth, it has been reported that Nedd4-1 conjugates mono-ubiquitination to GTP-bound Rap2A, which abolishes the Rap2A interaction with TRAF2 and NCK-interacting protein kinase (TNIK), so that Nedd4-1 acts as a positive regulator of dendrite growth (Kawabe et al., 2010). In addition to the function of Nedd4-1 in dendrite development, recent functional analyses of Nedd4-1 and Nedd4-2 double conditional knockout mice have revealed that Nedd4-1 and Nedd4-2 are also involved in axon growth in hippocampal neurons (Hsia et al., 2014). In terms of neuronal cell fate, Nedd4-1 has been reported to regulate dorsoventral patterning of the neuronal ectoderm in zebrafish by regulating ΔNp63α (Bakkers et al., 2005). In Drosophila, Nedd4-1 conjugates polyubiquitin chains to Notch and leads to its endocytosis, indicating that a regulatory function of Nedd4 in cell fate determination is conserved in Drosophila (Dalton et al., 2011; Sakata et al., 2004). In terms of neuronal cell survival, it has been reported that Nedd4-2 facilitates the ubiquitination of TrkA, leading to its trafficking to late endosomes upon NGF treatment in cultured dorsal root ganglion neurons (Yu et al., 2011, 2014).
While there has been significant interest in the functions of Nedd4-1 and Nedd4-2 in the developing brain, the functions of Nedd4-1 and Nedd4-2 in adult brain remain uncharted territory. Several biochemical studies have proposed that voltage-gated sodium channels (Nav1.2, Nav1.3, Nav1.5, and Nav1.7), voltage-gated potassium channels (KCQN2/3 and KCQN3/5), and voltage-dependent calcium channels (Cav1.2) are regulated by Nedd4 subfamily members, indicating possible roles of neuronal Nedd4-1 and Nedd4-2 in the regulations of neuronal excitability (Ekberg et al., 2007; Fotia et al., 2004; Rougier et al., 2005, 2011; Schuetz et al., 2008). In our study, we identified two novel substrate proteins for Nedd4-1 and Nedd4-2, Kir4.1, an inwardly rectifier channel expressed mainly in astrocytes and Connexin-43 one of the main gap junction proteins in astrocytes.
4.1.3 Kir4.1 is a substrate of Nedd4 subfamily E3 ligases in astrocytes
Kir4.1 is the main potassium channel in astrocytic end-feet at synapses (Robel and Sontheimer, 2015). Propagation of action potentials upon neuronal activity results in increases of the extracellular potassium concentration (Nicholson and Syková, 1998). During
76 hyperactivity, extracellular potassium concentrations have been reported to elevate from 3 mM to 10-12 mM (Heinemann and Lux, 1977). Considering that an increase in extracellular potassium leads to more positive membrane potentials, this affecting the activation of ion channels, transporters, and receptors, the clearance of excess potassium from extracellular space at axons and synapses strongly affects neuronal excitability (Seifert and Steinhäuser, 2013). It has been reported that brain-specific Kir4.1 conditional knockout mice show pronounced behavioral changes with ataxia and seizures (Djukic et al., 2007). In whole-genome linkage studies, homozygous missense mutations in KCNJ10 gene which encodes the Kir4.1 channel, were been reported as the cause of SeSAME syndrome (EAST Syndrome), which is characterized by seizures, ataxia, sensorineural deafness, mental retardation, and electrolyte imbalance (Bockenhauer et al., 2009; Scholl et al., 2009). Further, an impaired function of Kir channels has been reported in the CA1 region of sclerotic human epileptic hippocampal specimens (Kivi et al., 2000). Our data indicate that Nedd4-1 and Nedd4-2 conjugate K63-linked polyubiquitin chains to the C-terminal cytoplasmic tail of Kir4.1 and thus regulate Kir4.1 protein level in astrocytes (Figure 3-4 and 3-5), and thereby might play an important role in the regulation of neuronal excitability in adult brain through astrocytes.
In addition to potassium buffering, astrocytes play an important role in the clearance of glutamate from the synaptic cleft (Anderson and Swanson, 2000). Indeed, glutamate uptake and potassium buffering are interdependent processes. Astrocytic Kir4.1 plays an important role in establishing the negative membrane potential, providing the driving force of glutamate uptake by astrocytes (Wetherington et al., 2008). Correspondingly, downregulation or selective deletion of Kir4.1 in cultured astrocytes lead to inhibition of glutamate uptake by astrocytes (Djukic et al., 2007; Kucheryavykh et al., 2007). After uptake into astrocytes, glutamate is converted to glutamine by glutamine synthetase, and glutamine cycles back to the neurons through a process called glutamate-glutamine cycle (Danbolt, 2001). In neurons, glutamine can then be converted into glutamate or gamma-aminobutiric acid (GABA) (Robel and Sontheimer, 2015). Disruption of the glutamate-glutamine cycle by selective inhibitors of neuronal glutamine transporters or astrocytic glutamine synthetase significantly reduced evoked inhibitory postsynaptic currents (eIPSCs) in hippocampal CA1 pyramidal neurons, indicating that glutamate uptake by astrocytes and the glutamate-glutamine cycle has an important role in inhibitory synaptic transmission in the brain (Liang et al., 2006), while excitatory synaptic transmission can work
77 independently from the glutamate-glutamine cycle (Kam and Nicoll, 2007). Therefore, imbalanced glutamate uptake by astrocytes might affect the excitation/inhibition balance in neuronal networks (Robel and Sontheimer, 2015). Given that Nedd4-1 and Nedd4-2 regulate Kir4.1 levels in astrocytes, they might have a role in neuronal networks in terms of excitation/inhibition balance through astrocytes. This seems to be the case and roles of Nedd4-2 in neuronal excitability and excitation/inhibition balance in neuronal networks will be further discussed in section 4.1.5.
4.1.4 Nedd4-2 is the dominant E3 ligase regulating Connexin-43 in astrocytes
Nedd4-1 was identified as an E3 ligase for Connexin-43 by affinity purification using GST-tagged Connexin43 from a hepatocyte cell line (Leykauf et al., 2006). In the same report, the second WW domain of Nedd4-1 was identified as the Connexin-43-binding domain, and the PPXY motif in Connexin-43 was mapped as the Nedd4-1-binding site. Strikingly, knockdown of Nedd4-1 by siRNA was shown to cause an increase in membrane localization of Connexin-43 in a cultured hepatocyte cell line (Leykauf et al., 2006), and a follow-up study proposed that Nedd4-1 conjugates multiple single ubiquitin moieties to Connexin-43 and that ubiquitinated Connexin-43 interacts with ubiquitin interacting motif (UIM) of Eps15, leading the internalization of Connexin-43 (Girão et al., 2009).
As mentioned already, such knockdown studies using cultured cell lines are often problematic. Although we identified Connexin-43 as a protein that is upregulated in the Nedd4-1/2 bDKO as compared to Nedd4-Nedd4-1/2 bCtl in our proteome screening (Figure 3-5A), quantitative Western blotting in Figure 3-6 demonstrated that the Connexin-43 level is upregulated in the Nedd4-2 bKO but not in the Nedd4-1 bKO, indicating that Nedd4-2 is the dominant E3 ligase for Connexin-43 in the brain. Supporting this notion, the results of the in vivo ubiquitination assays demonstrated that Nedd4-2 has a stronger activity to conjugate K63-linked polyubiquitin chains to Connexin-43 than Nedd4-1 (Figure 3-5B-C). However, different degrees of upregulation of Connexin-43 levels in Nedd4-1/2 bDKO (1.95 fold) and Nedd4-2 bKO (1.55 fold) indicate that in the absence of Nedd4-2, Nedd4-1 might partially compensate the loss of function of Nedd4-2 (Figure 3-6A, 3-6C).
Astrocytes are coupled with each other through gap junctions and thereby form a large intercellular network in the brain that allows astrocytes to disperse ions and small molecules,
78 such as ATP, K+, Ca2+, glutamate, and cAMP (Dbouk et al., 2009; Wetherington et al., 2008).
Connexin-43 is one of the main gap junction proteins expressed in astrocytes. Dye-coupling experiments revealed that in Connexin-43 conditional knockout mice, the overall astrocytic intercellular networking is reduced by 50% (Theis et al., 2003a, 2003b). In a study with Connexin-30 and Connexin-43 double knockout mice, it was shown that astrocyte coupling through these connexins accelerates potassium buffering in hippocampal slices and that mice lacking Connexin-30 and Connexin-43 are more susceptible to epileptiform events (Wallraff, 2006). Astrocytic gap junction coupling is also important for ICWs in the astrocytic network and gap junction inhibitors reduce or block ICWs (Rouach and Giaume, 2001; Venance et al., 1995). In addition to gap junction coupling between neighboring astrocytes, one of the molecular mechanisms in the propagation of ICWs is the release of ATP through Connexin-43 hemichannels (Stout et al., 2002). Purinergic receptors are involved in ICWs by triggering the IP3-signaling pathway (Suadicani et al., 2004; Venance et al., 1997). Given that Nedd4-2 regulates the level of Connexin-43 in astrocytes, our data indicate that Nedd4-2 might play an important role in the regulation of ICWs in astrocytic networks, eventually affecting the function of neuronal networks.
Regulation of Connexin-43 in astrocytic network might also play an important role in potassium buffering (Wallraff, 2006), suggesting dual effects of Nedd4-2 on potassium clearance in the brain through both Kir4.1 and Connexin-43 in astrocytes.
4.1.5 Nedd4-1 and Nedd4-2 in astrocytes regulate neuronal network function through Kir4.1 and Connexin-43
Considering two critical roles of Nedd4-1- and Nedd4-2-dependent regulation of Kir4.1 and Connexin-43 in astrocytes, spatial potassium buffering and ICWs, we investigated neuronal functions in our knockout mouse lines. Our data on gamma oscillations (Figure 8E, 9E, 3-10E) indicate that the synchronization of neuronal function is impaired in the CA3 pyramidal region of the hippocampus in Nedd4-1/2 bDKO and Nedd4-2 bKO, but not in Nedd4-1/2 nDKO mice. This shows that glial Nedd4-1 and Nedd4-2 have an impact on neuronal function in the hippocampus. Imbalanced potassium clearance because of the increased level of Kir4.1 might disrupt the neuronal function in Nedd4-1/2 bDKO and Nedd4-2 bKO mice. The increased level of Kir4.1 might also affect the glutamate uptake by astrocytes leading to alterations in the
79 excitation/inhibition balance due to an imbalance in the glutamate-glutamine cycle. Impairments in gamma oscillations in Nedd4-1/2 bDKO and Nedd4-2 bKO might be due to alterations in excitation/inhibition balance in the neuronal network as it has been reported that the excitation/inhibition balance in the brain is important for proper neuronal network function (Mann and Mody, 2009; Snijders et al., 2013). Similarly, altered hippocampal synaptic inhibition in Neuroligin-4 knockout mice results in reduced gamma oscillation in hippocampal slices (Hammer et al., 2015), and kainate-induced gamma oscillations are highly dependent on synaptic inhibition by parvalbumin-expressing basket cells (Bartos et al., 2007).
Another reason of reduced gamma oscillatory activity in 1/2 bDKO and Nedd4-2 bKO might be the increased levels of Connexin-43 in astrocytes. It has been reported, for instance, that Connexin-43 overexpression alters the properties of ICWs (Suadicani et al., 2004) and astrocytic intracellular calcium responses precede carbachol-induced gamma oscilaltions in the CA3 region of the hippocampus (Lee et al., 2014). Thus, the increased levels of Connexin-43 in Nedd4-1/2 bDKO and Nedd4-2 bKO mice might perturb the properties of ICWs, thereby causing reduced gamma oscillatory activities in the CA3 region of the hippocampus.
Additionally, the increased levels of Connexin-43 might result in high copy numbers of Connexin-43 hemichannels in Nedd4-1/2 bDKO and Nedd4-2 bKO mice, which might lead to an increased levels of ATP-release by hemichannels. The increased levels of ATP-release by astrocytes might lead to changed properties of ICWs through purinergic receptors on neighbor astrocytes, thereby leading to reduced gamma oscillatory activities in Nedd4-1/2 bDKO and 2 bKO mice. Also, the increased levels of Connexin-43 in 1/2 bDKO and Nedd4-2 bKO mice, together with increased levels of Kir4.1, might affect the potassium buffering by astrocytes, which might lead to altered neuronal excitability, thereby affecting gamma oscillatory activities in the CA3 region of hippocampus. It would be particularly important to study if specific inhibitors of Kir4.1 and Connexin-43, such as fluoxetine hydrochloride and GAP-26 peptide respectively (Desplantez et al., 2012; Ohno et al., 2007), rescue the phenotypes of reduced gamma oscillatory activities in the CA3 region of the hippocampus in Nedd4-1/2 bDKO and Nedd4-2 bKO mice. This experiment would lead us to conclude direct impact of increased levels of Kir4.1 and/or Connexin-43 in Nedd4-1/2 bDKO and Nedd4-2 bKO mice on gamma oscillatory activities in CA3 region of hippocampus. It would be also particularly important to study if the increased levels of Kir4.1 and Connexin-43 have an impact on neuronal
80 excitability in Nedd4-1/2 bDKO and Nedd4-2 bKO by measuring the mEPSCs and mIPSCs in pyramidal neurons at the CA3 region of hippocampus in brain slices. This experiment would lead us to conclude if activities of excitatory or inhibitory neurons are altered in the CA3 region of the hippocampus of Nedd4-1/2 bDKO and Nedd4-2 bKO mice.
4.1.6 Regulation of Prr7 by Nedd4-2 might play a role in spine morphology
Prr7 is a postsynaptic protein composed of a short extracellular N-terminal region, a single transmembrane domain, and a C-terminal cytoplasmic tail containing several PPXY motifs (Hrdinka et al., 2011). Prr7 interacts with PSD95 and NMDA receptor subunits NR1 and NR2B (Murata et al., 2005). Although the function of Prr7 is unknown, a recent proteomics screen of ubiquitinated synaptic proteins has revealed that Prr7 is one of the postsynaptic proteins that are highly ubiquitinated in the rat brain (Na et al., 2012).
In the present study, we discovered that Nedd4-2 conjugates K63-linked polyubiquitin chains to Prr7 and thereby regulates the levels of Prr7 (Figure 3-11). Considering that Prr7 interacts with NMDA receptor subunits, Nedd4-2 might regulate the function or localization of NMDA receptors through Prr7 ubiquitination. Analyses of dendritic spines on CA1 pyramidal cells of Nedd4-2f/f-NEX-Cre+/- (Nedd4-2 nKO) and control mice have shown that in Nedd4-2 nKOs, the length of filipodia and mushroom spines are increased whereas spine density is not altered; additionally, the abundance of bifurcated spines is increased at the expense of mushroom spines (unpublished data from our group by Dr. Mateusz Cyryl Ambrożkiewicz).
These findings and the produced data on Prr7 indicate that Nedd4-2 might play a role in dendritic spine development by regulating Prr7. It would be particularly important to study if the knock-down of Prr7 rescues the phenotype of spine morphology in Nedd4-2 nKO mice. This experiment would lead us to conclude the direct impact of increased levels of Prr7 on spine development.